Large-scale and long-term biodiversity monitoring can be improved by the use of Essential Biodiversity Variables, among which species population sizes provide key data for conservation programs. Relevant estimations and assessment of actual population sizes are critical for species conservation, especially in the current context of global biodiversity erosion. However, knowledge on population size varies greatly, depending on species conservation status and ranges. While the most threatened or restricted-range species generally benefit from exhaustive counts and surveys, monitoring efficiently common and widespread species tends to be neglected or is simply more challenging to achieve. In such a context, citizen science (CS) is a powerful tool for the long-term monitoring of common species through the engagement of various volunteers, permitting data acquisition on the long term and over large spatial scales. Despite this substantially increased sampling effort, detectability issues imply that even common species may remain unnoticed at suitable sites. The use of structured CS schemes, including repeated visits, enables to model the detection process, permitting reliable inferences of population size estimates. Here, we relied on a large French structured CS scheme (EPOC-ODF) comprising 27 156 complete checklists over 3 873 sites collected during the 2021-2023 breeding seasons to estimate the population size of 63 common bird species using Hierarchical Distance Sampling (HDS). These population size estimates were compared to the previous expert-based French breeding bird atlas estimations, which did not account for detectability issues. Our results indicate strong under-estimations for 65% of species in the French atlas, likely due to more conservative estimates inferred from semi-quantitative expert-based assessments. We also found a few over-estimations for species with long-range songs such as Cuculus canorus, Upupa epops or Turdus merula. Our study highlights the need to rely on sound statistical methodology to ensure unbiased ecological inferences with adequate uncertainty estimation and advocates for a higher reliance on structured CS in support of biodiversity monitoring.